firecrawl 4.0.0__tar.gz → 4.1.1__tar.gz

{firecrawl-4.0.0 → firecrawl-4.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: firecrawl
-Version: 4.0.0
+Version: 4.1.1
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/firecrawl/firecrawl
 Author: Mendable.ai

{firecrawl-4.0.0 → firecrawl-4.1.1}/firecrawl/__init__.py

@@ -17,7 +17,7 @@ from .v1 import (
     V1ChangeTrackingOptions,
 )
 
-__version__ = "4.0.0"
+__version__ = "4.1.1"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

{firecrawl-4.0.0 → firecrawl-4.1.1}/firecrawl/v2/client.py

@@ -18,6 +18,7 @@ from .types import (
     CrawlResponse,
     CrawlJob,
     CrawlParamsRequest,
+    PDFParser,
     CrawlParamsData,
     WebhookConfig,
     CrawlErrorsResponse,
@@ -105,7 +106,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,
@@ -571,7 +572,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,
@@ -759,7 +760,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,

{firecrawl-4.0.0 → firecrawl-4.1.1}/firecrawl/v2/types.py

@@ -278,7 +278,7 @@ class ScrapeOptions(BaseModel):
     timeout: Optional[int] = None
     wait_for: Optional[int] = None
     mobile: Optional[bool] = None
-    parsers: Optional[List[str]] = None
+    parsers: Optional[Union[List[str], List[Union[str, 'PDFParser']]]] = None
     actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None
     location: Optional['Location'] = None
     skip_tls_verification: Optional[bool] = None
@@ -536,6 +536,11 @@ class PDFAction(BaseModel):
     landscape: Optional[bool] = None
     scale: Optional[float] = None
 
+class PDFParser(BaseModel):
+    """PDF parser configuration with optional page limit."""
+    type: Literal["pdf"] = "pdf"
+    max_pages: Optional[int] = None
+
 # Location types
 class Location(BaseModel):
     """Location configuration for scraping."""
@@ -594,6 +599,8 @@ class SearchRequest(BaseModel):
         
         return normalized_categories
 
+    # NOTE: parsers validation does not belong on SearchRequest; it is part of ScrapeOptions.
+
 class LinkResult(BaseModel):
     """A generic link result with optional metadata (used by search and map)."""
     url: str
@@ -686,4 +693,4 @@ AnyResponse = Union[
     MapResponse,
     SearchResponse,
     ErrorResponse,
-]
+]

{firecrawl-4.0.0 → firecrawl-4.1.1}/firecrawl/v2/utils/validation.py

@@ -311,6 +311,20 @@ def prepare_scrape_options(options: Optional[ScrapeOptions]) -> Optional[Dict[st
                                 converted_action[action_key] = action_value
                         converted_actions.append(converted_action)
                 scrape_data["actions"] = converted_actions
+            elif key == "parsers":
+                converted_parsers = []
+                for parser in value:
+                    if isinstance(parser, str):
+                        converted_parsers.append(parser)
+                    elif isinstance(parser, dict):
+                        converted_parsers.append(parser)
+                    else:
+                        parser_data = parser.model_dump(exclude_none=True)
+                        # Convert snake_case to camelCase for API
+                        if "max_pages" in parser_data:
+                            parser_data["maxPages"] = parser_data.pop("max_pages")
+                        converted_parsers.append(parser_data)
+                scrape_data["parsers"] = converted_parsers
             elif key == "location":
                 # Handle location conversion
                 if isinstance(value, dict):
@@ -321,4 +335,4 @@ def prepare_scrape_options(options: Optional[ScrapeOptions]) -> Optional[Dict[st
                 # For fields that don't need conversion, use as-is
                 scrape_data[key] = value
     
-    return scrape_data 
+    return scrape_data  

{firecrawl-4.0.0 → firecrawl-4.1.1}/firecrawl.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: firecrawl
-Version: 4.0.0
+Version: 4.1.1
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/firecrawl/firecrawl
 Author: Mendable.ai